linux-kernel - RE: [GIT PULL] pid: use flex array

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8b73a39a359b4523a053efd614cd7dd7@AcuMS.aculab.com>
Date:   Fri, 30 Jun 2023 11:21:44 +0000
From:   David Laight <David.Laight@...LAB.COM>
To:     'Christian Brauner' <brauner@...nel.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>
CC:     Kees Cook <keescook@...omium.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [GIT PULL] pid: use flex array

From: Christian Brauner
> Sent: 30 June 2023 09:04
...
> > And some of them are even ok. I don't think it's always wrong,
> > particularly if you then abstract it out.
> >
> > So doing something like that
> >
> >    #define PCI_IOBASE ((void __iomem *)0)
> >
> > makes perfect sense. It's literally abstracting out something real (in
> > this case yes, it looks like a NULL pointer, but it's actually a
> > pointer with a strict type that just happens to have the value zero.

You can't do that, it doesn't work.
A NULL pointer is any constant integer expression that evaluates
to zero implicitly or explicitly converted to a pointer type.

The bit pattern used for a NULL pointer is implementation defined.
It is almost always 0, but the C language allows any invalid
value to be used - eg the 'all ones' pattern.

clang warns for 'PCI_IOBASE + 4'.
Probably because it only has the expected value if NULL is the
zero bit pattern - so it isn't portable.

If has to be said that I doubt gcc or clang support NULL being
other than the zero bit pattern.
Any C code that uses memset() to set a pointer to NULL is making
that assumption.

I have used a system where the native 'invalid pointer' was ~0
but the C port used zero.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)